So it's similar with "collect it all" approach?
Maybe; it's difficult to say. I think that governments around the world would prefer that we all use CBDCs.
I can give a reasonably plausible scenario: Perhaps the Bitcoin economy will be allowed to grow — with HW using compromised chips being standardised for KM — while the dollar/Euro/Yen/RMB economy declines. Then at some point in the future the United Nations announces that all BTC transactions are outlawed for [insert superficial reason here] and all BTC will be converted at parity with CBCD. Any address associated with affected HW/keys (majority) then has its BTC automatically drained into a new CBDC account. The minority not affected are offered amnesty. Such a "big bang" event would be difficult to oppose or counter given the massive additional power and leverage CBDCs would give any government over individuals. You may say this is a bit far-fetched - but it's plausible.
I think i get your point. But IMO it's not plausible, especially about revealing government have access or backdoor to people's devices with ability to steal or take away people's asset.
In terms of safeguarding/solutions:
1. Can we say that using non-specific generalized chips to store keys is safer as it's more decentralized and harder to backdoor than specialized secure elements? I believe Trezor takes this line.
2. Generating keys off-device by dice/coins throws and calculating checksum word using offline air gapped hardware that never connect to any network ever?
3. Using air gapped wallets for signing?
4. Developing subliminal free signatures so that it can be proven that keys are not leaked?
5. DIY HW projects using generalized off-the-shelf components?
I feel it's trade-off between lower security in general and less likely to contain specific backdoor.
2. Generating keys off-device by dice/coins throws and calculating checksum word using offline air gapped hardware that never connect to any network ever?
3. Using air gapped wallets for signing?
5. DIY HW projects using generalized off-the-shelf components?
There are already some project and people using these approach. But it's less convenient and require higher technical skill.
4. Developing subliminal free signatures so that it can be proven that keys are not leaked?
I barely understand what is subliminal free signatures. But,
1. Using software which choose nonce or K value carefully could avoid private key leak/recovery. RFC6979 handle such task.
2. Not re-using address means you avoid such concern.